BRENNEN

Industries / Food, farming and forestry / Food and beverage processing / Brewing and distilling

Quality and consistency

Quality and consistency is the work of making each batch taste as intended and safe to drink, through purity rules, a shared tasting vocabulary, lab tests for spoilage microbes and checks that spirits are what their labels say.

6

Milestones achieved, from 1516 to 2024.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1516

    Barley, hops and water only

    On 23 April 1516 in Ingolstadt, the Bavarian dukes Wilhelm IV and Ludwig X issued an ordinance that beer be brewed from barley, hops and water alone, and brewers who knowingly broke it had their barrels taken (Library of Congress). The German Brewers' Federation gives its aims as keeping beer prices down, keeping wheat for bread and stopping brewers from adding cheap and often poisonous plants. A German imperial tax law of 1906 applied the rule beyond Bavaria and listed yeast as an ingredient of bottom-fermented beers, meaning lagers. Bavarian law still bans sugar and spices in beer brewed there (LGL, Bavaria's food safety authority).

    LessonIn 1987 the European Court of Justice ruled that Germany could not use the rule to keep out beer lawfully brewed in other member states, finding the ban out of proportion to any health aim; German brewers still follow it.

    lgl.bayern.deblogs.loc.govbrauer-bund.deeur-lex.europa.eu

    Credit

    Dukes Wilhelm IV and Ludwig X of Bavaria

  2. 1979

    A shared language for flavor

    From 1974 to 1979 a joint working group of the American Society of Brewing Chemists (ASBC), the European Brewery Convention (EBC) and the Master Brewers Association of the Americas (MBAA) built the beer flavor wheel, a chart that groups 122 flavor notes into 14 classes (Oxford Companion to Beer). Morten Meilgaard, C. E. Dalgliesh and J. F. Clapperton published it as "Beer Flavor Terminology" in 1979. It gives brewers in different countries the same words for what they taste, and is used with reference chemicals added to beer to train tasting panels.

    beerandbrewing.comapi.crossref.org

    Credit

    Morten Meilgaard, C. E. Dalgliesh and J. F. Clapperton; ASBC, EBC and MBAA

  3. 1997

    Hop resistance traced to genes

    Bitter acids from hops keep most bacteria from growing in beer, but a few kinds resist them. In 1997 M. Sami, H. Yamashita and colleagues found a gene, horA, on a plasmid (a small loop of DNA) in hop-resistant Lactobacillus brevis, a major cause of spoiled beer. In 2001 K. Sakamoto, W. N. Konings and colleagues showed that HorA is a pump that pushes hop compounds and drugs out of the cell. These genes became targets for PCR, a test that copies and detects a stretch of DNA, to tell beer spoilers from harmless bacteria. Oldham and Held (2023) describe PCR kits sold to breweries and running costs of $1 to $3 a sample.

    pmc.ncbi.nlm.nih.govebi.ac.ukpmc.ncbi.nlm.nih.gov

    Credit

    M. Sami, H. Yamashita and colleagues; K. Sakamoto, W. N. Konings and colleagues

  4. 2012

    Methanol in Czech spirits

    Methanol is a toxic alcohol. Fermentation makes small amounts of it, and some counterfeiters add it to spirits to raise profits. In 2012 and 2013 the Czech Republic had a mass outbreak of methanol poisoning: of 121 people with confirmed poisoning, 20 died outside hospital and 21 in hospital, a death rate of 34% (Zakharov and colleagues, Clinical Toxicology, 2014). Outbreaks recur. The Pan American Health Organization (PAHO) reported 225 suspected cases in Brazil between August and 5 October 2025, 16 of them confirmed, and 15 deaths.

    LessonPAHO ties such outbreaks to adulterated drinks and spirits made outside official systems, and urges countries to watch for cases and control those drinks.

    ebi.ac.ukpaho.orgebi.ac.uk

    Credit

    documented by Zakharov and colleagues; PAHO

  5. 2018

    Bomb carbon unmasks fake whisky

    Nuclear weapons tests in the 1950s and early 1960s raised the level of radiocarbon, a slightly radioactive form of carbon, in the air. Barley took it up as it grew, so the level in a whisky dates its grain. Testing for the firm Rare Whisky 101, the Scottish Universities Environmental Research Centre (SUERC) found that 21 of 55 rare Scotch whiskies bought on the secondary market were outright fakes or not distilled in the year on the label, including a bottle sold as Ardbeg 1885. SUERC says it can place whisky distilled after the 1950s within two to three years.

    gla.ac.uk

    Credit

    SUERC (Gordon Cook, Elaine Dunbar); Rare Whisky 101 (David Robertson, Andy Simpson)

  6. 2024

    Pocket methanol test validated

    Researchers at ETH Zurich built a handheld detector that separates methanol from ethanol in the vapor above a drink and measures it with a small gas sensor (van den Broek and colleagues, Nature Communications, 2019). In a 2024 trial run to the ISO 5725 standard for comparing labs, 17 independent participants from six countries, among them distilleries, universities and authorities, made 119 measurements. The detector measured methanol down to 0.01% by volume within 2 minutes when used by non-specialists, with results that varied by less than 9.5% between labs. It is sold as the Alivion Spark M-20.

    ebi.ac.ukpmc.ncbi.nlm.nih.gov

    Credit

    Jan van den Broek, Andreas Güntner, Sotiris Pratsinis and colleagues; Alivion

Watch for

  1. Germs in alcohol-free draft beer

    Researchers at the University of Nottingham sequenced the microbes in more than 50 samples of 20 no- and low-alcohol beers poured from taps in 12 pubs in Nottingham, UK. They report that Acetobacter, Brettanomyces and Saccharomyces were core members of the microbe community, and that the pub a beer came from explained 72.6% of the variation between samples. They call it the first study to use long-read nanopore sequencing on these beers, and urge strict cleaning of lines, taps, keg couplers and nozzles for such "microbiologically sensitive" drinks. Diageo helped support the work.

    Achieved whenanother group finds the same pattern in pubs elsewhere, or a trial shows that a cleaning rule for these taps lowers microbe counts.

    pmc.ncbi.nlm.nih.gov

    Claimed by

    Roselli, Smart, Crow and Powell, October 2026

  2. Telling beer types by infrared

    Near-infrared spectroscopy shines infrared light through a sample and reads which wavelengths it absorbs. Wei and colleagues, at Heilongjiang Bayi Agricultural University and three other Chinese universities, report a deep-learning model on such spectra that predicted alcohol content with an R² (a measure of fit where 1 is perfect) of 0.995 on an independent test set. It also sorted craft beer, industrial beer and a lab-made unfermented imitation with 100% accuracy. Their reference alcohol values came from bottle labels, and the test set was split off from the same collection of samples.

    Achieved whenanother lab tests the method on beers bought separately, with alcohol measured by a standard lab method, and gets similar accuracy.

    pmc.ncbi.nlm.nih.gov

    Claimed by

    Wei and colleagues, Molecules, March 2026

Open

  1. 01

    Spoilage genes across many breweries

    A public map of beer-spoiling bacteria on brewery surfaces across many breweries. Swabs from fillers, keg washers, heat exchangers, barrels and draft lines are tested by qPCR (quantitative PCR, which counts copies of a DNA sequence) for the hop-resistance genes horA, horB, horC and hitA, and the counts are published by surface type.

    The brief
    Credit · unclaimed
  2. 02

    Flavor models on new beers

    An outside test of the KU Leuven beer-flavor models on beers they have never seen. Measure the models' main chemical inputs in about 30 commercial beers that were not among the original 250, predict their flavor scores, and compare the predictions with scores from an independent trained tasting panel.

    The brief
    Credit · unclaimed

Open 01 · the brief

Spoilage genes across many breweries

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public map of beer-spoiling bacteria on brewery surfaces across many breweries. Swabs from fillers, keg washers, heat exchangers, barrels and draft lines are tested by qPCR (quantitative PCR, which counts copies of a DNA sequence) for the hop-resistance genes horA, horB, horC and hitA, and the counts are published by surface type.

Why it is still open

Bokulich and colleagues swabbed one North American brewery in 2015 and called their results "the first indication of hop-resistance-gene abundance within a brewery environment". Other studies screen isolates taken from beers, not brewery surfaces. Searches of Europe PMC in November 2026 found no study that repeats the gene counts across many breweries. Oldham and Held (2023) write that breweries of all sizes need routine, affordable detection of spoilage microbes.

pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov

Built with what exists

The primer and probe sequences in Table 1 of the Bokulich paper; a benchtop qPCR machine; DNA extraction kits named by Oldham and Held, such as Bio-Rad InstaGene Matrix and Qiagen DNeasy PowerSoil Pro; each brewery's written consent; anonymized publication.

  1. First step

    Ask three local breweries for permission, and run the horA test on swabs from one filler and one draft line at each.

  2. Done looks like

    20 licensed breweries, at least ten surfaces each, sampled twice, with gene copies per square centimeter published by surface type and compared with each brewery's own culture results.

Open 02 · the brief

Flavor models on new beers

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An outside test of the KU Leuven beer-flavor models on beers they have never seen. Measure the models' main chemical inputs in about 30 commercial beers that were not among the original 250, predict their flavor scores, and compare the predictions with scores from an independent trained tasting panel.

Why it is still open

Schreurs and colleagues (Nature Communications, 2024) measured 226 chemical properties in 250 beers and trained machine-learning models on them. Their gradient-boosting models reached R² values (a measure of fit where 1 is perfect) of up to 0.75 for sensory features, but were tested on beers held back at random from the same 250. Of the 63 papers that Europe PMC lists as citing the study in November 2026, none, judged by title, runs the published models on new beers.

pmc.ncbi.nlm.nih.govebi.ac.uk

Built with what exists

The authors' code, data and trained models on Zenodo (record 10653704, CC BY 4.0); a contract lab with gas chromatography–mass spectrometry for aroma compounds; a trained panel at a brewing school or licensed brewery; Python and scikit-learn.

  1. First step

    Download the Zenodo record, rerun the published notebook, and send the 15 most important inputs from the paper's ranking to two contract labs for a quote.

  2. Done looks like

    For 30 new beers, the R² for each predicted attribute, published beside the paper's own test-set R², with all new data released.